With so many people spending a lot of time stuck in traffic, it is really one of the things that need to be resolved. Not only is it wasted time, but it also contributes to significant fuel consumption that can affect the environment.
The good news is, there could be a solution to this and if successful, we may no longer worry about traffic congestion.
Artificial Intelligence (AI) can help reduce traffic congestion through the recent research of Aston University. The AI is the first of its kind that can scan live footage and adjust the lights to compensate to keep traffic moving and reduce congestion on the road.
The researchers used deep reinforcement learning wherein the software can recognize when it's not doing well. When it does, it will attempt a new approach. On the other hand, if things are going well, the system will continue to improve to make further progress.
Also Read: MIT Researchers Used AI to Make Traffic Go Smoothly and Reduce Fuel Consumption and Emissions
How Did They Do It?
Aston University researchers built a state-of-the-art photo-realistic traffic simulator, the Traffic 3D. This stimulator trained their program and taught it to handle various traffic and weather scenarios.
It was tested on a real junction and it adapted to real traffic intersections despite being trained on simulations. Therefore, it could be effective in the real world.
"We have set this up as a traffic control game. The program gets a 'rewarded' when it gets a car through a junction. Every time a car has to wait or there's a jam, there's a negative reward. There's actually no input from us; we simply control the reward system," said Dr. Maria Chili from Aston University.
Currently, the main form of traffic light automation used at junctions relies on the magnetic induction loops wherein a wire sits on the road and registers cars passing over it. The program counts that and then reacts to the data.
The AI can see high traffic volume before the cars go through the lights, and it makes its decision.
The researchers based the program on learned behaviors to better understand situations that it hasn't experienced before. It was also tested with a physical obstacle that causes congestion, which the system did well.
The program can be set up to view any traffic junction, whether simulated or real, and will learn autonomously. Moreover, the reward system can be manipulated to allow it to let emergency vehicles through quickly. However, the program will always teach itself instead of being programmed with specific instructions.
The researchers plan to begin the test of the system on real roads this year. Should it be successful, there could really be a significant decrease in traffic congestion on many roads all over the world.
Related Article: Is Artificial Intelligence Really A Threat To Humanity?
This article is owned by TechTimes
Written by April Fowell
